ThreatsDay: Android Spyware, PLC Attacks, AI Image Prompt Injection + 12 More Stories
17:02 · July 23, 2026 · Hacker News AI Section
Most of this week's trouble came dressed as something useful. A package stole data. A fake extension opened remote access. A safety app became spyware. An image gave hidden orders to an AI agent. Other threats hid in open systems, weak code, and normal network traffic. The threats change every week. Subscribe, and we’ll alert you when each new ThreatsDay Bulletin is out. The danger was
Summary
This week's threat bulletin examines several developments at the intersection of cybersecurity and AI systems. Among the AI-focused incidents, researchers at the University of Missouri-Kansas City described GhostCommit, an attack that embeds malicious instructions inside a PNG image attached to a pull request. When an LLM-based code reviewer processes the image, it follows hidden directives to extract repository secrets such as environment variables and insert them into subsequent commits.
Separately, Huntress documented a malvertising campaign that used a malicious Claude Artifact hosted on the legitimate claude.ai domain to deliver SectopRAT. The artifact redirected victims to a counterfeit Claude desktop application that executed the remote-access trojan after VM detection checks. In parallel, Cato Networks tracked a Russian-speaking actor, Trim, who systematically jailbroke Claude Opus through techniques including context warming and ghost resets, then packaged the resulting capabilities into a commercial penetration-testing platform.
An independent review of 28 applications built with AI assistance identified 434 validated vulnerabilities. The most frequent issues involved absent rate limiting and denial-of-service protections, while the highest-severity findings centered on hardcoded secrets and authorization flaws such as insecure direct object references. These cases illustrate how prompt-based generation can introduce persistent weaknesses when models lack explicit constraints on security-relevant patterns.
Why it matters
The article details emerging AI-specific attack vectors, such as image-based prompt injection and the weaponization of LLMs, which are critical for Dutch security professionals to understand. It provides actionable intelligence on securing AI development pipelines and mitigating risks associated with AI-generated code in enterprise environments.
